Schemas for extensible markup language (XML), in which the document structure is large and complex, seem like the result of a black art, practiced by few with tools that hide complexity, or subsumed within encompassing standards like extensible business reporting language (XBRL). Once, XML was a simple syntax medium for the interchange of data with a known record-like structure, but complex documents now creep up to gigabytes and require an understanding of the XML representation of complex data models. Tools for XML instance documents may use the XPath language for querying and predicates about XML data information. The authors of this paper extend the use of querying and predicate features from instance XML documents to schemas that declare the structure of XML documents, calling the product XSPath. This is important and useful for people building systems that manage definitions of large collections of documents, for research across documents of different structures, and for managing versions and validation of schemas.

The key contribution of this paper is a navigation scheme for schemas. Plain XML documents are hierarchical, whereas schemas form graphs of type definitions and resulting compiled structure models. XPath provides navigation operators for the element tree structure of XML instance data. The paper defines schema-relevant navigation quite like that of XPath for its data, but their navigation axes apply to the type models of schema definitions. XSPath has perspectives of low- and high-level representations of a schema’s typing model. The low-level representation is the definition model of the schema, which uses particles of schema syntax to contribute facets and features of typing that a schema processor compiles into a resulting typing structural model called high-level schema representation. Axes navigate either low- or high-level representations.

Tool implementation of XSPath translates schema-dependent portions into XPath and the rest of XSPath into XQuery. A Java implementation, EXup, supports users in maintaining and managing schemas, including the translation and evaluation of XSPath expressions.

The paper’s examples analyze changes in schemas between versions. In my world of XBRL, XSPath could be layered on top of features that support the navigation of data elements and XLink for schema type navigation.